Design of an agent server for neonatal analysis and trend detection

A Neonatal Intensive Care Unit (NICU) generates vast amounts of physiological data, much of which is not recorded. This physiological data may contain information to enable early warning of possible complications in neonatal babies conditions. An agent-based intelligent decision support system (IDSS) for neonatal analysis and trend detection is being developed in order to examine this physiological data. This paper specifically looks at what features are be provided by an agent server as part of an agent-based IDSS framework for neonatal analysis and trend detection. To relieve the agents of added complexity, the Agent Server not only provides assistance with communication but other additional supplementary features. First the paper presents related work already performed including the high level framework of which this system is a component. Then the paper overviews the Agent Server, including some of the features. The key features of the agent server presented in this paper are the communication method used in order to simplify agent communication, the database access method used to simplify database access to multiple databases of different types by the agents, and the finally the method for the dynamic allocation of agents to tasks including instantiating of additional agents if necessary.